From ix-flow
Run an Agent IX workflow with ix-flow — create a run and drive it through its phases, pausing at human gates for approval. Use when asked to run, start, continue, or resume an ix-flow workflow.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ix-flow:ix-flowThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Drive an `ix-flow` workflow to completion. The user names a workflow (and may point at a
Drive an ix-flow workflow to completion. The user names a workflow (and may point at a
skill directory); you create a run and advance it through its phases, stopping at human
gates for their approval.
Pass --json to every command and read the result envelope — state, current_phase,
open_gates[].token, and next_actions tell you what to do next.
Resolve the workflow. Use a registered definition name directly, or a skill
directory with --path <skill-dir>. If a run is already in progress, skip to step 3
with its run id.
Create the run.
ix-flow run <workflow> [--path <skill-dir>] --json
Record instance_id (the run id) and current_phase.
Advance through the phases. Read the workflow's own SKILL.md for how to proceed,
then move along its declared transitions:
ix-flow status <run-id> --json
ix-flow advance <run-id> <next-phase> --json
If state is invariant_failed, the run stayed put — satisfy what the failure reports,
then advance again.
Handle human gates. When advance returns state: gate_deferred, the workflow is
waiting on human approval. Present what is being approved to the user and wait for their
decision. Once they approve, record it and continue:
ix-flow ack <run-id> <token> --reviewer <user>
ix-flow advance <run-id> <next-phase> --json
The <token> is open_gates[0].token from the deferred result.
Finish. Stop when the run reaches a terminal phase. Report the final phase and a short summary of what happened.
~/.ix/flows; use ix-flow resume <run-id> to pick one back up in a
new session.Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub agent-ix/ix-flow --plugin ix-flow